Understanding the Local Epidemiology of Methicillin-Resistant Staphylococcus aureus by Pulsed-Field Gel Electrophoresis in an Indian Tertiary Care Centre

Background: Methicillin-resistant Staphylococcus aureus (MRSA) is an important cause of community and hospital-acquired infections (HAIs). In the year preceding this study, our laboratory reported an MRSA isolation rate of 2% from 50,549 specimens. Molecular typing of MRSA identies sources of infection, transmission chains and informs infection control practices, and pulsed-eld gel electrophoresis (PFGE) is the gold standard. This study was conducted to gain an understanding of the local molecular epidemiology of MRSA in our hospital using PFGE, to inform hospital infection control practices. Methods: This prospective longitudinal study was conducted in the microbiology laboratory of our 2,200-bed tertiary care teaching hospital in Mumbai, India. The antibiotic susceptibility proles and pulsed-eld proles (PFPs) of 100 consecutive non-duplicate clinical isolates of MRSA were obtained. The PFPs were compared to check for relatedness of isolates. The distribution of various pulsotypes across disciplines and hospital locations was examined. Results: Clinical specimens accounted for 86 (86%) of the MRSA isolates, whilst 14 (14%) were from screening of healthcare workers. Maximum isolates, 68 (68%), were from surgical disciplines. Conrmed HAIs accounted for 25 (25%) MRSA isolates. Seventeen antibiotypes were obtained and there was no correlation between antibiotype and pulsotype. Totally 43 pulsotypes were identied, with most isolates, 40 (40%), belonging to pulsotype 1. Seven clusters were identied. Cluster I had maximum pulsotypes, 14, and 58 (58%) isolates. Isolates belonging to clusters I and II were found in all hospital locations. Relatedness was observed between isolates from HAIs and screening specimens, and between community and HAI isolates. Conclusions: PFGE typing revealed the disciplines at greatest risk from MRSA in our hospital. The commonality between MRSA isolated from HAIs and screening of healthcare workers, and between MRSA isolated from HAIs and from community-acquired infections highlighted the horizontal transmission of MRSA and the need to reinforce infection control measures to limit this.


Background
Methicillin-resistant Staphylococcus aureus (MRSA) causes illnesses that are di cult to treat as they are resistant to β-lactam antibiotics and, frequently, multiple other antimicrobials [1,2]. Selection pressure arising from the inappropriate use of antibiotics has led to MRSA becoming a concern in healthcare settings and in the community [1,3]. Its propensity to colonise human skin, form surface bio lms, and transmissibility through direct skin contact or via fomites make MRSA an important cause of hospitalacquired infections (HAIs) [1,2,4,5].
Typing of MRSA is central to controlling its spread, enabling early detection of outbreaks, identifying sources of infection, transmission chains and informing infection control practices [6,7]. Pulsed-eld gel electrophoresis (PFGE) is a band-based genotyping method. The chromosomal DNA of MRSA is digested with the rare-cutting restriction enzyme SmaI derived from Serratia marcescens. This enzyme cleaves speci cally at CCC^GGG sites on the chromosome, yielding DNA fragments that range in size from 10 kilobases -800 kb [8]. The digested DNA fragments are separated by agarose gel electrophoresis in an electric eld with an alternating voltage gradient. The resultant banding patterns produced by different isolates are compared to determine their relatedness [6,8]. The advantages of PFGE that make it the gold standard for epidemiological typing of MRSA are: its high discriminatory index (> 0.95), concordance of greater than 95% with sequence-based genotypic methods, viz., multilocus sequence typing (MLST) and Staphylococcal protein A (spa) typing, and lower cost [6,9].
India is a high prevalence country for MRSA, with a prevalence of 37% (21%-45%) [10]. It is estimated that 55%-65% of S. aureus isolates from India are resistant to cefoxitin [11]. Data on MRSA carrier rates vary widely from 1.8% among healthy community-dwelling population to 23% among healthcare workers (HCWs). The epidemiology and transmission of MRSA in Indian healthcare settings has only been sparsely studied at a molecular level [12][13][14][15].
The aim of this study was to characterise MRSA from clinical isolates in our centre by PFGE pro les (pulsotypes), examine the relatedness of various isolates, and determine the distribution of MRSA pulsotypes across hospital locations. This was done to understand the local epidemiology of MRSA at our centre at the molecular level. This understanding would inform the development and implementation of hospital infection control policies tailored to our unique situation.

Methods
Study setting, design and duration: This prospective longitudinal study was carried out in the microbiology laboratory of a 2,200-bed tertiary care teaching hospital in Mumbai, India over 18 months. In the year preceding this study, our laboratory reported an MRSA isolation rate of 2% from 50,549 specimens processed for bacteriology. Disciplines from which MRSA was predominantly isolated (72% of all MRSA) were general surgery, paediatric surgery, burns unit and orthopaedics. Our department also periodically screens HCWs for MRSA carriage.
De nitions of terms used: Isolate: Pure culture of MRSA derived from a single colony of a single organism [8].
Cluster: A group of isolates that produced PFGE banding patterns that were more than 80% similar, and that could be distinguished from other isolates on this basis [8,17,18].
Singleton: Isolate with a unique PFP not belonging to any of the identi ed clusters [17,18].
Sample size, inclusion and exclusion criteria: One hundred (100) consecutive non-duplicate isolates of MRSA from specimens received in the microbiology laboratory during the study period were included. Isolates of MRSA obtained from HCWs who underwent routine screening as part of hospital infection control protocol during the study period were also included. Screening was not carried out exclusively for the purpose of this study.
Repeated MRSA isolates: (i) from the same site from the same patient isolated during the same course of hospitalisation and, (ii) in case of outpatients, within two weeks of previous isolation in our laboratory, were excluded.
All specimens were processed according to standard protocols for isolation, identi cation and antimicrobial susceptibility testing of MRSA [19,20].

Typing of MRSA isolates by PFGE:
The Centres for Disease Control's PulseNet protocol was followed and is brie y described below [21].
Four or ve colonies of the test isolate were inoculated into 5 ml brain heart infusion broth (HiMedia, India), vortexed and incubated at 35°C for 24 h. After incubation, the concentration of each cell suspension was adjusted with sterile normal saline to an absorbance of 0.9 to 1.1 at 610 nm using a spectrophotometer (JLab, India).
Two hundred microlitres of the adjusted cell suspensions were centrifuged at 10,000 rpm for six minutes.
The pellets were resuspended in 300 µl of Tris-EDTA (TE) buffer (ThermoFisher, USA) and equilibrated in a 37°C dry bath. Cell lysis was carried out by adding 4 µl conventional lysostaphin solution (Sigma-Aldrich, USA) to each tube. The tubes were incubated in a 37°C dry bath for an hour till the solutions had cleared.
After incubation, 300 μl of 1.8% w/v PFGE grade agarose (Sisco Research Laboratories, India) equilibrated to 55°C was added to the cell suspensions in each tube, mixed gently and 100 µl each dispensed into the wells of a small disposable plug mould (Bio-Rad, USA). The plugs were solidi ed at room temperature and incubated in EC lysis buffer overnight at 37°C.
The plugs were then washed in 4 ml TE buffer by agitating the tubes on an orbital rocker for 30 min after decanting the EC lysis buffer. The wash cycle was repeated four more times, using fresh TE buffer each time. The plugs were stored in TE buffer at 2°C -8°C till use.
The plugs were removed from TE buffer and each cut to a size of 5 mm x 5 mm using a sterile blade to t a 15-tooth comb. These were equilibrated at room temperature for 30 min in 200 µl of a 1:10 mixture of 10X CutSmart restriction buffer (New England BioLabs, USA) and Type I water. Following this, all the buffer-water mixture was aspirated from each tube and the plugs incubated overnight at 25°C in a 200 µl mixture of SmaI restriction enzyme (New England BioLabs, USA) in 1:10 10X CutSmart restriction buffer and Type I water.
A 1% w/v PFGE grade agarose gel was prepared in 0.5X Tris-borate-EDTA (TBE) buffer (ThermoFisher, USA). Twelve plugs were included in each PFGE run along with three lambda reference standards [22]. Plug slices were loaded at the ends of the comb teeth. Reference standards of Lambda PFG Ladder (New England BioLabs, USA) of 1 mm thickness were sliced from the gel syringe and loaded in the rst, middle (8 th ) and last (15 th ) teeth of the comb. Agarose equilibrated to 55°C was poured into the gel casting platform and allowed to solidify at room temperature for one hour.
After completion of the run, the gel was stained in a 1.5 g/ml ethidium bromide solution (HiMedia, India) for 20 min in a covered container and de-stained in fresh distilled water for 45 min. The gel was documented using a Syngene gel doc system (Synoptics, UK) and the image saved in TIFF format with a minimum resolution of 768 x 640 pixels for further analysis [21,23].

Data analysis:
Antimicrobial susceptibility patterns were analysed for the phenotypic relatedness of isolates. Isolates with susceptibility differing for two or fewer drugs of the same class (not including cefoxitin) were classi ed in a single antibiotype.
Pulsed-eld pro les of the isolates were compared by visual assessment based on the criteria (Table 1) proposed by Tenover et al [8]. These criteria were applied only if PFGE resolved at least 10 distinct bands per isolate. In case fewer than 10 bands were resolved on two repeat runs, or smearing was observed on two repeat runs (one with 0.5% SDS added prior to loading to free up the DNA from restriction enzyme), the isolate was classi ed as untypeable by PFGE [24].
A dendrogram was created to identify percent similarities. This was derived from the unweighted pair group method using arithmetic averages and Dice coe cients [23]. The various MRSA pulsotypes were mapped across hospital locations.

Results
The 100 non-duplicate consecutive MRSA isolates studied were obtained from 52 (52%) male and 48 (48%) female subjects. The median age of subjects from which MRSA was isolated was 28 y (range 1 d -65 y).
Of the 100 MRSA isolates, 14 (14%) were obtained from HCWs from whom it was isolated during screening. Eleven were from clinical staff of obstetrics and gynaecology and three from general surgery.
The relationship between isolates obtained from HAIs and screening from various disciplines is shown in Figure 4.
In cluster I, 29 MRSA isolates from community-acquired infections were closely related to HAI and screening isolates. Similar observations were made from clusters II (six isolates from communityacquired infections), IV (one isolate) and VII (one isolate) (Fig. 3).
The clinical departments in our hospital are distributed between four buildings within an area of approximately 10 acres. Building A houses the departments of dermatology, internal medicine, general surgery, obstetrics and gynaecology, the burns unit, paediatrics, and paediatric surgery. Building B houses neonatology, gastroenterology and gastrointestinal surgery, neurology, neurosurgery, haematology and the laboratories. Building C houses orthopaedics, and Building D the cardiology, cardiothoracic surgery and pulmonology departments. The distribution of the various clusters between hospital buildings is shown in Figure 5 (screening isolates were not included in this gure as these were considered to be dispersed throughout the hospital).

Discussion
Of the 100 MRSA isolates studied, 87% were from the surgical disciplines of general surgery, orthopaedics, paediatric surgery and obstetrics and gynaecology. Skin and soft tissue infections (SSTIs) were the commonest presentation (48%). Maximum isolates (76%) were from wound swabs and pus specimens. Community infections accounted for 47% of our isolates. S. aureus is known to be the leading cause of SSTIs [10,25]. Since a majority of our isolates were from SSTIs for which therapy required surgical intervention, most of our isolates were obtained from specimens submitted by the surgical disciplines.
The antimicrobial susceptibility patterns of MRSA isolates in our study agree with current trends in India and globally [10,26]. However, resistance to aminoglycosides, tetracycline, clindamycin and trimethoprimsulfamethoxazole was infrequently encountered. The latter three are not commonly used in our hospital. Interestingly, although aminoglycosides are frequently used, resistance was infrequent.
Maximum isolates belonged to antibiotype Ab2 (Table 2), and were obtained from all hospital locations. Expectedly, there was no correlation between antibiotype and pulsotype as antibiotic susceptibility patterns depend on antibiotic usage [16,28]. For example, antibiotype Ab2 had isolates belonging to such diverse clusters I, II, IV, V, VI, as well as several singletons (Fig. 3).
The rst report from India on the molecular typing of MRSA using PFGE was by Gayatri, et al. in 2005 [14]. Genotyping of 82 single-patient isolates of MRSA from two major hospitals revealed four major types and 24 subtypes. Also, isolates that belonged to the same strain by MLST and spa typing were found to be diverse by PFGE. Relatedness was observed to the Hungarian strain HUSA304 and the Brazilian strain HSJ216 that were circulating at the time [14]. In 2010, PFGE established the presence of EMRSA-15 variants in India in hospital and community in urban and rural settings, and grouped all pulsotypes into a single cluster as they were more than 80% similar [15]. We did not include any known MRSA strains in our PFGE experiments as the primary aim of our study was to understand the local epidemiology of MRSA in our hospital in terms of sources and transmission.
In our study, 99 MRSA isolates resolved into 43 pulsotypes belonging to seven clusters and there were 16 singletons. Cluster I was predominant with the maximum number of pulsotypes (14) and isolates (58%) (Fig. 3). Similar studies examining the molecular epidemiology of MRSA in hospital settings also reveal considerable molecular diversity of circulating MRSA but nearly always a single strain predominates. Jain, et al. typed 46 MRSA isolates from infected orthopaedic implants, grouping them into nine clusters based on a similarity cut-off of 80%, with one cluster being predominant. In their study, maximum isolates from across PFGE clusters were found by MLST to belong to the same clonal complex, indicating considerable diversity within members of the same complex [17]. Fomda, et al. assigned 154 MRSA isolates from their tertiary care centre into ten pulsotypes, suggesting less diversity than that found in our centre. However, the origins of their isolates with regard to hospital or community are not elucidated [29]. Krishnan, et al. also assigned 65 MRSA isolates from burns wounds to 12 pulsotypes, with one pulsotype predominating [30].
We isolated MRSA belonging to cluster I from all four of our hospital buildings (Fig. 5). This was followed by isolates belonging to cluster II which were found in three of four buildings. Building A, the design and layout for which date back to 1926, houses the general clinical disciplines with approximately 900 beds in total, and is the most crowded. This building accounted for both maximum number and diversity of MRSA. This was followed by Building C, which houses orthopaedics and has 200 beds. Orthopaedics accounted for a considerable number of MRSA isolates (15%) as well as diversity in our study. Buildings B and D, which have a modern layout and design, house the super-specialty departments and are considerably less crowded, accounted for less number and diversity of MRSA (Fig. 5).
Carriage of MRSA by HCWs is a proven risk factor for HAIs [28,31]. In this study, isolates obtained from HAIs and screening specimens were, for the most part, related (Fig. 4). For example, cluster I comprised isolates from three SSIs from obstetrics and gynaecology, and general surgery, as well as nine screening isolates form the same disciplines. Interestingly, we isolated an MRSA belonging to cluster I from an infected copper-T that had been inserted several years ago in another hospital in the same city indicating possible widespread circulation of this strain in our city.
As many as 20% of HCWs in a hospital may carry MRSA at any point in time [31,32]. The genetic relatedness of screening isolates and isolates from HAIs in our study (Fig. 4) illustrates how failure to comply with infection control practices results in the transmission of MRSA between HCWs and patients. Dar, et al. also showed by PFGE the genetic relatedness of MRSA isolated from orthopaedic SSIs and screening specimens [28].
In clusters I, II, IV and VII, MRSA isolates from community-acquired infections were found to be closely related to isolates from HAIs (Fig. 3). Attempts at controlling MRSA within a hospital setting have been known to be hampered by the repeated in ux of MRSA strains from the community [33]. Preventing this in ux and also preventing the e ux of MRSA from the hospital into the community demands stringent adherence to infection control practices as well as antimicrobial stewardship to prevent the emergence of resistance in the rst place [34]. Screening inpatients on admission and decolonisation of carriers has been shown to be successful in controlling the spread of MRSA [35,36].
A single isolate in our study was deemed untypeable by PFGE as band separation could not be achieved. A possible explanation could be that cleavage by the restriction endonuclease was blocked by methylation of the bacterial DNA. Members of MRSA ST398 are untypeable by SmaI PFGE due to the presence of a novel DNA methyltransferase [37].
Our study has a few limitations. Screening HCWs for MRSA was not carried out for the purpose of this study. Only MRSA isolates obtained from screening of disciplines carried out routinely as part of hospital infection control protocol during the study period were included. Environmental sampling was not carried out. It is therefore possible that some sources of MRSA may have been missed. Secondly, inter-laboratory comparisons and identi cation of our isolates with known epidemic strains of MRSA could not be carried out as these were not included in our PFGE runs [6]. Sequence-based typing by SCCmec, MLST and spa would yield this information and place our ndings in the broader context of national and global epidemiology of MRSA, providing a complete macro-epidemiological picture.
Nevertheless, this was the rst time a study of the molecular epidemiology of MRSA in our centre was undertaken. This study highlighted the disciplines at greatest risk from MRSA transmission, viz., the surgical disciplines. Intensifying surveillance and control measures on these disciplines would limit transmission. The study revealed the commonality between MRSA isolated from HAIs and screening specimens, impressing the need to reinforce basic infection control measures, such as hand hygiene and audits, in areas at maximum risk. The commonality observed between MRSA isolated from HAIs and those from community-acquired infections suggested the need to screen admitted patients for MRSA carriage and to decolonise carriers (at least in areas at greatest risk), as well as reinforce basic infection control measures and antimicrobial stewardship. Finally, a protocol for the gold-standard PFGE typing was standardised and successfully employed that could also be used in future epidemiological investigations.
This study of the molecular epidemiology of MRSA by PFGE in our hospital thus served as an important tool for informing infection control practices and enabling targeted prevention of HAIs with economical utilisation of resources.

Conclusions
Typing by PFGE revealed the genetic diversity of MRSA in our centre. Isolates belonging predominantly to two clusters were found circulating throughout our hospital. Isolates from HAIs and screening specimens were related, suggesting horizontal transmission between HCWs and patients. Commonality was also observed between hospital and community isolates, suggesting possible introduction into the hospital from the community or vice versa. The datasets used and/or analysed during the present study are available from the corresponding author on reasonable request. Tables   Table 1 Criteria for interpreting PFGE patterns [8].

Category
Number of Fragment Differences Similarity Index
*Intermediate susceptibility to erythromycin. # Intermediate susceptibility to vancomycin. Figure 1 Antimicrobial susceptibility patterns of MRSA isolates.

Figure 2
Antimicrobial susceptibility patterns of MRSA isolates.